Why OpenAI may well be completely Zuck’d

Regular readers of this Substack will recall that I have never been bullish on OpenAI. Last Fall, I warned that they might someday be known as the WeWork of OpenAI, and in late January I wrote an essay about some of the strong headwinds they faced, despite their seeming invincibi … | Continue reading


@garymarcus.substack.com | 5 months ago

Don’t look up: the massive Microsoft/Crowdstrike data outage is a huge wake-up call

Wake up. If a single bug can take down airlines, banks, retailers, media outlets, and more, what on earth makes you think we are ready for AGI? The world needs to up its software game massively. We need to invest in improving software reliability and methodology, not rushing out … | Continue reading


@garymarcus.substack.com | 5 months ago

Marcus goes gaga over Gates clip

It’s been a long while since I have felt really positive about most of what I have been reading and seeing in AI. Not because I inherently dislike AI, but because I feel like so many folks have been brainwashed by the “scale is all you need” notion and “AGI in 2025” (or 2027 or 2 … | Continue reading


@garymarcus.substack.com | 6 months ago

The need for a President that speaks AI natively

Last night was a travesty, but that is just the beginning of our problems | Continue reading


@garymarcus.substack.com | 6 months ago

Clarification from Ray Kurzweil

He’s still down for 2029, same as ever | Continue reading


@garymarcus.substack.com | 6 months ago

GPT-5… now arriving Gate 8, Gate 9, Gate 10

The increasingly delayed countdown to GPT5 | Continue reading


@garymarcus.substack.com | 6 months ago

The Great AI Retrenchment has begun

Further evidence that AGI is not imminent | Continue reading


@garymarcus.substack.com | 6 months ago

The misguided backlash against California’s SB-1047

State Senator Scott Wiener and others in California have proposed a bill, SB-1047m that would some modest (to my taste) restraints on AI. It doesn’t call for a private right of action, which would allow individual citizens to sue AI companies for a wide set of reasons; it doesn’t … | Continue reading


@garymarcus.substack.com | 7 months ago

AGI by 2027?

Fun with charts | Continue reading


@garymarcus.substack.com | 7 months ago

First sign of GenAI Winter?

Investors appear to have taken note | Continue reading


@garymarcus.substack.com | 7 months ago

Open letter responding to Yann LeCun

A memo for future intellectual historians | Continue reading


@garymarcus.substack.com | 7 months ago

“OpenAI Insiders Warn of a ‘Reckless’ Race for Dominance”

Passing along this scoop from Kevin Roose: Roose supplied a gift link: https://x.com/kevinroose/status/1797992577255518480?s=61 The letter itself, cosigned by Bengio, Hinton, and Russell, can be found here https://righttowarn.ai. I fully endorse its four recommendations: | Continue reading


@garymarcus.substack.com | 7 months ago

AI as Reality Television

Perhaps no week of AI drama will ever match the week in which Sam got fired and rehired, but the writers for the AI reality series we are all watching just don’t quit. For one thing, the bad press about Sam Altman and OpenAI, who once seemingly could do no wrong, just keep coming … | Continue reading


@garymarcus.substack.com | 7 months ago

26 months of ridicule and failure

The wall is reliability. | Continue reading


@garymarcus.substack.com | 7 months ago

Disturbing and misleading efforts to defame Helen Toner through misdirection

OpenAI’s new board just made its loyalties clear. We should all be worried/ | Continue reading


@garymarcus.substack.com | 7 months ago

Kara Swisher, Sam Altman, and the OpenAI Board

Helen Toner finally explains what the board was thinking | Continue reading


@garymarcus.substack.com | 7 months ago

Elon Musk: 1; Yann LeCun: 1; Humanity: 0

In case you haven’t heard, there‘s a brawl happening over at X, Musk vs LeCun. For days I tried to resist commenting, but so many people (friends, reporters, etc) keep asking me for my opinion I have decided to oblige. The first thing that I will tell you is that each of those no … | Continue reading


@garymarcus.substack.com | 7 months ago

Sam Altman, then and now

Update to my last essay, “What should we learn from OpenAI’s mistakes and broken promises?” Sam Altman, 2016: “We’re planning a way to allow wide swaths of the world to elect representatives to a new governance board. Because if I weren’t in on this I’d be, like, Why do these fuc … | Continue reading


@garymarcus.substack.com | 7 months ago

What should we learn from OpenAI’s mistakes and broken promises?

It’s increasingly clear that OpenAI has not been consistently candid. What follows from that? | Continue reading


@garymarcus.substack.com | 7 months ago

ScarJo is just the tip of the iceberg

As recently as November 2023, OpenAI promised in their filing as a nonprofit exempt from income tax to make AI that that “benefits humanity … unconstrained by a need to generate financial return”. The first step towards that should be a question about product – are the products w … | Continue reading


@garymarcus.substack.com | 7 months ago

Partial Regurgitation and how LLMs really work

A sticky example / Up with which I will not chuck | Continue reading


@garymarcus.substack.com | 7 months ago

Sam Altman’s fall from grace

Sam Altman is trending and virtually every comment is trashing him. | Continue reading


@garymarcus.substack.com | 7 months ago

AI has become Sauron’s Ring

Another of AI’s bitter lessons | Continue reading


@garymarcus.substack.com | 7 months ago

The OpenAI board was right

Even on little things, Sam is not consistently candid. | Continue reading


@garymarcus.substack.com | 7 months ago

Jane Goodall, AI Skeptic

Your loyal correspondent goes gaga for Goodall | Continue reading


@garymarcus.substack.com | 7 months ago

Chaos and tension at OpenAI

Safety seems to be taking a back seat | Continue reading


@garymarcus.substack.com | 7 months ago

Hot take on OpenAI’s new GPT-4o

GPT-4o hot take: • The speech synthesis is terrific, reminds me of Google Duplex (which never took off). but • If OpenAI had GPT-5, they have would shown it. • They don’t have GPT-5 after 14 months of trying. • The most important figure in the blogpost is attached below. And the … | Continue reading


@garymarcus.substack.com | 7 months ago

Microsoft and OpenAI’s increasingly complicated relationship

An AI Soap Opera in the making? | Continue reading


@garymarcus.substack.com | 8 months ago

The Sam Altman Playbook

Fear, The Denial of Uncertainties, and Hype | Continue reading


@garymarcus.substack.com | 8 months ago

An epistemic clusterfuck in the making

Social media was bad. Adding AI into the mix could easily get a lot worse. | Continue reading


@garymarcus.substack.com | 8 months ago

We can’t trust the fox to guard the henhouse, especially when it comes to AI

The backstory behind Taming Silicon Valley | Continue reading


@garymarcus.substack.com | 8 months ago

Humans versus Machines: The Hallucination Edition

So many people are confused about the relation between human cognitive errors and LLM hallucinations that I wrote this short explainer: Humans say things that aren't true for many different reasons • Sometimes they lie • Sometimes they misremember things | Continue reading


@garymarcus.substack.com | 8 months ago

Facing Facts

“The definition of insanity is doing the same thing over and over again while expecting different results.” If all we had was ChatGPT, we could say, hmm “maybe hallucinations are just a bug”, and fantasize that they weren’t hard to fix. If all we had was Gemini, we could say, hmm … | Continue reading


@garymarcus.substack.com | 8 months ago

Groupthink versus Tulips. What will historians think of 2024?

Some thoughts occasioned by Meta’s new model and a bad week in the stock market | Continue reading


@garymarcus.substack.com | 8 months ago

Daniel Dennett, 1942-2024

A remembrance from Doug Hofstadter | Continue reading


@garymarcus.substack.com | 8 months ago

Technoptimism, TED, and The Road to the Future

It’s kind of surreal to compare some of the talks at TED yesterday with reality. Microsoft’s Mustafa Suleyman promised that hallucinations would be cured “soon”, yet my X feed is still filled with examples like these from Princeton Professor Aleksandra Korolova: | Continue reading


@garymarcus.substack.com | 8 months ago

TED, AI, and the test of time

How well did last year’s talks hold up? | Continue reading


@garymarcus.substack.com | 8 months ago

LLMs aren’t very bright. Why are so many people fooled?

How Generative AI plays on human cognitive vulnerability | Continue reading


@garymarcus.substack.com | 8 months ago

Evidence that LLMs are reaching a point of diminishing returns — and what that might mean

The conventional wisdom, well captured recently by Ethan Mollick, is that LLMs are advancing exponentially. A few days ago, in very popular blog post, Mollick claimed that “the current best estimates of the rate of improvement in Large Language models show capabilities doubling e … | Continue reading


@garymarcus.substack.com | 8 months ago

Superhuman AGI is not nigh

Why Elon was probably wise not to take the bet | Continue reading


@garymarcus.substack.com | 8 months ago

$10 million says we won’t see human-superior AGI by the end of 2025

An open letter to Elon Musk | Continue reading


@garymarcus.substack.com | 9 months ago

24 hours in AI news

A bunch of small but important items in recently-breaking AI news: Tesla settled their lawsuit with the Huang family over Walter Huang’s death, for an undisclosed amount of money. “Although Huang’s family acknowledges he was distracted while the car was driving, they argued Tesla … | Continue reading


@garymarcus.substack.com | 9 months ago

Corrected link, re: new paper on scaling and AGI

With apologies, there was a broken link in my last post (now corrected in the online version of the post). The important new arXiv article can be found here https://arxiv.org/abs/2404.04125. Thank you all for your support! –Gary | Continue reading


@garymarcus.substack.com | 9 months ago

Breaking news: Scaling will never get us to AGI

A new result casts serious doubt on the viability of scaling | Continue reading


@garymarcus.substack.com | 9 months ago

Generative AI as Shakespearean tragedy

You may have read yesterday’s New York Times report by Cade Metz and others on how many of the biggest AI companies have been cutting ethical corners in a race to gather as much data as possible (“OpenAI, Google and Meta ignored corporate policies, altered their own rules and dis … | Continue reading


@garymarcus.substack.com | 9 months ago

Information pollution reaches new heights

AI is making shit up, and thar made-up stuff is trending on X | Continue reading


@garymarcus.substack.com | 9 months ago

An unending array of jailbreaking attacks could be be the death of LLMs

A couple days ago I reported a survey saying that most IT professional are worried about the security of LLMs. They have every right to be. There seems to be an endless number of ways of attacking them. In my forthcoming book, Taming Silicon Valley, I describe two examples. The f … | Continue reading


@garymarcus.substack.com | 9 months ago

Sneak preview of GPT-5!

Mind blown. | Continue reading


@garymarcus.substack.com | 9 months ago